I'm a full-stack engineer who goes looking for the hard problems:
- 🔥
$\color{#E5534B}{\textsf{the outage nobody can explain}}$ - 📈
$\color{#D18616}{\textsf{the queue that backs up at peak traffic}}$ - 🏗️
$\color{#539BF5}{\textsf{the platform rebuilt while production keeps running}}$
🧩 Backend and infrastructure are my core, and I ship the frontend too. That lets me take a system from the first design doc to the dashboard that shows it holding up under load.
💻 Backend
Services that stay fast and correct under real load.
Go, Python, TypeScript, Rust, FastAPI, Node.js, Celery
🎨 Frontend
Internal tools with a UI people want to use.
React, TypeScript, Vite
☁️ Cloud and infrastructure as code
Every resource lives in git, and every change goes through a pull request.
AWS, GCP, Azure, Terraform, AWS CDK, Atlantis, Cloud Custodian
☸️ Kubernetes and delivery
Clusters that scale themselves and deploy straight from git.
EKS, GKE, Karpenter, Cilium, Istio, KEDA, Argo CD, Helm, Kustomize, Harbor
📨 Data and messaging
Queues and databases that absorb traffic spikes without losing work.
PostgreSQL, RDS, RabbitMQ, Kafka, Elasticsearch, Redis
🔭 Observability
Follow any request from a log line to a full trace.
Quickwit, OpenTelemetry, Vector, VictoriaMetrics, Prometheus, Grafana
🔐 Security and compliance
Least privilege by default, with an audit trail on every change.
Kyverno, Keycloak, Teleport, Infisical, IRSA, SonarQube, PCI DSS, SOC 2, ISO 27001
- If it can't be rebuilt from git, it isn't done.
- Platforms are products: sensible defaults, paved paths, documentation engineers read.
- Fix the root cause once, across every service, rather than patching one incident at a time.
- Measure before optimizing, and right-size from real usage, not guesses.
- macOS-vz-kubelet: extending the virtual kubelet with a fuller Kubernetes feature set and Apple container runtime support for sidecars
- Running Quickwit in production for logs and traces, and following Tantivy and Pixie upstream
📫 Find me on X / Twitter